Menlo Research has unveiled the open source Asimov v1 humanoid robot with public CAD files, simulation models and onboard software, aiming to lower barriers for robotics labs, startups and independent builders developing next-generation humanoid systems.
Menlo Research has introduced the Asimov v1 humanoid robot as an open-source humanoid platform designed for builders, researchers and robotics developers, positioning humanoid robotics away from closed proprietary systems and towards reproducible engineering platforms.
Unlike many humanoid robots limited to polished demonstrations, Asimov v1 openly exposes its mechanical CAD, electrical CAD, MuJoCo simulation model, onboard software and bill of materials, allowing developers to build, train, modify, simulate and self-source the robot. Reported licensing includes CERN OHL-S 2.0 for hardware elements and GPL 2.0-style licensing for software components.
The 1.2 metre, 35 kg humanoid uses 25 actuated degrees of freedom alongside two compliant passive toe joints designed to improve ground contact, walking balance and locomotion stability while reducing software control burden. The platform delivers up to 120 Nm peak torque and uses a Raspberry Pi 5 alongside a Radxa CM5 compute stack.
Its sensor suite includes a monocular camera, quad microphone array, IMUs, speaker output and joint state feedback for perception and locomotion experimentation.
Menlo Research describes Asimov v1 as a platform users can “build, train and customize.”
The project also includes a MuJoCo simulation environment supporting sim-to-real locomotion research, reinforcement learning and processor-in-the-loop testing.
A DIY kit with a USD 499 deposit and a target price of around USD 15,000 is planned for summer 2026, alongside a fully self-sourced build route for advanced developers and robotics labs.
